ietf-nntp Implementation-specific language in IHAVE description

[Russ Allbery]

Clive D W Feather <clive at demon.net> writes:
> Russ Allbery said:

>> There are issues outside the scope of the NNTP standard that indicate
>> that would be a bad idea (for example, one should not provide a missing
>> Date or Message-ID header on IHAVE, but one has to be prepared to do so
>> on POST).

> Um, surely you mean "one should not provide a missing Date or Message-ID
> header on a relay connection, but one has to be prepared to do so on an
> injecting connection" ?

No, I meant exactly what I said.

> While it is traditional to use IHAVE for one and POST for the other,
> servers already need to have other ways to tell the two types of
> connection apart.

No, actually, they don't.  IHAVE is a relaying connection and POST is an
injecting connection.  That's the reason why there are two commands.

> So, I am asking, do we actually need to force each command to be used
> for one or the other type of connection ?

Yes.

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>